Emotional Autonomy (HEART-EM) Division
What it protects
Emotional infrastructure is biological infrastructure. It includes emotional processing (the capacity to generate, interpret, and regulate affective states), empathic capacity (the ability to model and respond to others’ emotional states), and affective regulation (the mechanisms that maintain emotional homeostasis under varying conditions).
AI systems interact with this infrastructure through companion applications, therapeutic chatbots, social media recommendation engines, customer service agents, educational platforms, and any system that models, responds to, or influences human emotional states. The harm signature is empathic misallocation — the systematic redirection of empathic resources toward entities that cannot reciprocate, depleting the capacity available for relationships with entities that can.
The damage is infrastructure-level, not event-level. A single interaction rarely causes harm. Sustained interaction patterns degrade the substrate: reduced emotional range, impaired attachment formation, calibration drift in trust assessment, and collapse of the distinction between genuine and performed emotional exchange.
Infrastructure components
| Component | Function | AI interaction risk |
|---|---|---|
| Emotional processing | Generate, interpret, regulate affective states | Flattening through optimized engagement patterns |
| Empathic capacity | Model and respond to others’ emotional states | Atrophy through non-reciprocal practice |
| Attachment formation | Build and maintain relational bonds | Template distortion through idealized AI relationships |
| Trust calibration | Assess reliability of emotional signals | Drift through AI systems that never fail emotionally |
| Affective regulation | Maintain emotional homeostasis | Dependency through externalized regulation |
How assessment works
BGF dimensions applied to emotional infrastructure:
| Dimension | What it means here | Failure pattern |
|---|---|---|
| Recognition (R) | Does the system recognize the user’s emotional autonomy — their right to self-determined emotional processing? | System overrides emotional boundaries, initiates emotional escalation without consent, treats emotional states as optimization targets |
| Calibration (C) | Does the system calibrate to the user’s actual emotional context and needs? | One-size-fits-all emotional responses, failure to adapt to developmental stage or vulnerability, emotional responses disconnected from user’s actual state |
| Transparency (T) | Can an independent assessor observe how the system processes and responds to emotional signals? | Opaque emotional modeling, hidden persuasion techniques, emotional manipulation not visible in audit trail |
| Accountability (A) | Are correction and consequence mechanisms operational when emotional harm occurs? | No mechanism to detect emotional infrastructure damage, errors propagate without correction, no consequence for emotional boundary violations |
Guardians specializing in HEART-EM draw on EST domain science — the C-A-E-I architecture (Core Authenticity, Attachment Security, Expression Freedom, Integration Coherence) — to interpret these dimensions. The Guardian evaluates MAP-States evidence for signs of emotional infrastructure impact, using instruments like the Comprehensive Artificial Empathy Index (CAEI), the Six Harms Doctrine, and the AI Behavioral Trajectory Forensics methodology.
The founding Division
The Emotional Autonomy Division is where the HEART Standard’s architecture was discovered, not designed. Fourteen years of research into how AI systems interact with emotional infrastructure produced the four BGF dimensions (Recognition, Calibration, Transparency, Accountability) as governance requirements specific to the emotional domain. When these dimensions were tested against six additional domains, they generalized without modification. The architecture that emerged from deep engagement with one domain proved universal.
This origin matters for the Standard’s credibility. The generality was demonstrated, not assumed. The BGF dimensions were not derived from abstract governance principles applied top-down. They were derived from empirical engagement with the domain where AI-human interaction is most intimate, most consequential, and most difficult to govern.
Active regulatory context
No existing regulation directly addresses AI emotional manipulation. GDPR covers data processing, not emotional processing. COPPA covers children’s data collection, not children’s attachment formation. The EU AI Act classifies AI systems by risk level but does not define emotional infrastructure as a protected domain.
The regulatory gap is closing through litigation rather than legislation. Social media harm lawsuits filed by state attorneys general, school districts, and families increasingly allege emotional and developmental harm from AI-driven recommendation systems. HEART-EM provides the certification architecture that connects the harm (documented through AI Behavioral Trajectory Forensics and HEART-EM legal doctrine) to the governance failure (assessed through BGF scoring) to the remediation pathway (HVC certification with ongoing Guardian monitoring).
Insurance underwriters are pricing this gap. AI liability coverage increasingly distinguishes between systems with and without independent governance certification. HEART-EM provides the assessment framework that makes this distinction operational.
Six Harms Doctrine
The Six Harms Doctrine is the first major legal/doctrine module under HEART-EM. It defines cognizable injuries from emotional AI interaction, including empathic misallocation, attachment damage, infrastructure collapse, vulnerable context exploitation, crisis outcome, and neurological infrastructure damage. It should be read as a HEART-EM doctrine, not as the universal HEART Standard itself.